Advice on Ph.D.-level Research

In September 2002, while the memories of my own graduate school experience at Cornell were still fresh, and I was just starting out my career as an Assistant Professor at USC, I wrote this document offering advice to new Ph.D. students in my group.

I had nearly forgotten all about it. I stumbled across it this morning, and thought I'd share it here.

Experience has taught me that it's hard to give one-size-fits-all advice, but much of what I wrote then still seems reasonable to me. One sentence I would certainly change in the document, though, is "Always, always, focus your research on problems not tools..." I'm not entirely sure what I was thinking of when I wrote this. Although I have been generally more problem-oriented in most of my own research on wireless networks, in recent years I've gained much respect for research on developing and enhancing theoretical tools that can address a wide range of problems. For instance, in ongoing work, we have developed an algorithm for linear combinatorial optimization with unknown random variables. I'd consider this one of the most satisfying pieces of research I've worked on, precisely because it provides a theoretical tool that is broadly applicable to problems in many disciplines beyond my own.